Implement streaming mode in ReorderBuffer.

Instead of serializing the transaction to disk after reaching the
logical_decoding_work_mem limit in memory, we consume the changes we have
in memory and invoke stream API methods added by commit 45fdc9738b.
However, sometimes if we have incomplete toast or speculative insert we
spill to the disk because we can't generate the complete tuple and stream.
And, as soon as we get the complete tuple we stream the transaction
including the serialized changes.

We can do this incremental processing thanks to having assignments
(associating subxact with toplevel xacts) in WAL right away, and
thanks to logging the invalidation messages at each command end. These
features are added by commits 0bead9af48 and c55040ccd0 respectively.

Now that we can stream in-progress transactions, the concurrent aborts
may cause failures when the output plugin consults catalogs (both system
and user-defined).

We handle such failures by returning ERRCODE_TRANSACTION_ROLLBACK
sqlerrcode from system table scan APIs to the backend or WALSender
decoding a specific uncommitted transaction. The decoding logic on the
receipt of such a sqlerrcode aborts the decoding of the current
transaction and continue with the decoding of other transactions.

We have ReorderBufferTXN pointer in each ReorderBufferChange by which we
know which xact it belongs to.  The output plugin can use this to decide
which changes to discard in case of stream_abort_cb (e.g. when a subxact
gets discarded).

We also provide a new option via SQL APIs to fetch the changes being
streamed.

Author: Dilip Kumar, Tomas Vondra, Amit Kapila, Nikhil Sontakke
Reviewed-by: Amit Kapila, Kuntal Ghosh, Ajin Cherian
Tested-by: Neha Sharma, Mahendra Singh Thalor and Ajin Cherian
Discussion: https://postgr.es/m/688b0b7f-2f6c-d827-c27b-216a8e3ea700@2ndquadrant.com

1 files changed, 37 insertions, 5 deletions

diff --git a/src/backend/access/heap/heapam_visibility.c b/src/backend/access/heap/heapam_visibility.c
index dba10890aab..c77128087cf 100644
--- a/src/backend/access/heap/heapam_visibility.c
+++ b/src/backend/access/heap/heapam_visibility.c
@@ -1571,8 +1571,25 @@ HeapTupleSatisfiesHistoricMVCC(HeapTuple htup, Snapshot snapshot,
 												 htup, buffer,
 												 &cmin, &cmax);
 
+		/*
+		 * If we haven't resolved the combocid to cmin/cmax, that means we
+		 * have not decoded the combocid yet. That means the cmin is
+		 * definitely in the future, and we're not supposed to see the tuple
+		 * yet.
+		 *
+		 * XXX This only applies to decoding of in-progress transactions. In
+		 * regular logical decoding we only execute this code at commit time,
+		 * at which point we should have seen all relevant combocids. So
+		 * ideally, we should error out in this case but in practice, this
+		 * won't happen. If we are too worried about this then we can add an
+		 * elog inside ResolveCminCmaxDuringDecoding.
+		 *
+		 * XXX For the streaming case, we can track the largest combocid
+		 * assigned, and error out based on this (when unable to resolve
+		 * combocid below that observed maximum value).
+		 */
 		if (!resolved)
-			elog(ERROR, "could not resolve cmin/cmax of catalog tuple");
+			return false;
 
 		Assert(cmin != InvalidCommandId);
 
@@ -1642,10 +1659,25 @@ HeapTupleSatisfiesHistoricMVCC(HeapTuple htup, Snapshot snapshot,
 												 htup, buffer,
 												 &cmin, &cmax);
 
-		if (!resolved)
-			elog(ERROR, "could not resolve combocid to cmax");
-
-		Assert(cmax != InvalidCommandId);
+		/*
+		 * If we haven't resolved the combocid to cmin/cmax, that means we
+		 * have not decoded the combocid yet. That means the cmax is
+		 * definitely in the future, and we're still supposed to see the
+		 * tuple.
+		 *
+		 * XXX This only applies to decoding of in-progress transactions. In
+		 * regular logical decoding we only execute this code at commit time,
+		 * at which point we should have seen all relevant combocids. So
+		 * ideally, we should error out in this case but in practice, this
+		 * won't happen. If we are too worried about this then we can add an
+		 * elog inside ResolveCminCmaxDuringDecoding.
+		 *
+		 * XXX For the streaming case, we can track the largest combocid
+		 * assigned, and error out based on this (when unable to resolve
+		 * combocid below that observed maximum value).
+		 */
+		if (!resolved || cmax == InvalidCommandId)
+			return true;
 
 		if (cmax >= snapshot->curcid)
 			return true;		/* deleted after scan started */